Expert Elicitation To Estimate the Feed Safety Impact of Criteria Included in the Canadian Food Inspection Agency Risk Assessment Model for Feed Mills.

Abstract: The Canadian Food Inspection Agency is developing an Establishment-based Risk Assessment (ERA) model for commercial and on-farm mills involved in the manufacture, storage, packaging, labeling, or distribution of livestock feed (ERA-Feed Mill model). This model will help inform the allocation of inspection resources on the basis of feed safety risk, including animal health and food safety risk. In a previous study, 34 risk factors, grouped into inherent, mitigation, and compliance clusters, along with assessment criteria were selected. The objective of this current study was to estimate the relative risk (RR) of the 203 assessment criteria on the basis of the impact on feed safety to design an ERA-Feed Mill model algorithm. Furthermore, the intent of this study was to assess the maximum increase or decrease of risk obtained when multiple criteria belonging to a same cluster were identified in a specific feed mill. To do so, a two-round face-to-face expert elicitation was conducted with 28 Canadian feed experts. Results showed no significant association between respondent profiles (years of experience and work sector) and estimated RR. Uniformity of answers between experts improved between rounds. Criteria having the highest increase in risk (median RR ≥ 4) included the presence of materials prohibited to be fed to ruminants in a facility that produces ruminant feed, the presence of multiple livestock species on-site, and historical noncompliances related to the inspection of the feed mill's process control and end-product control programs. Risk mitigation criteria having the highest impact on decreasing the risk were the implementation of feed safety certifications, the use of dedicated manufacturing lines (prohibited materials or medications), and having a hazard sampling plan in place for finished feed. The median RR assigned to each criterion and cluster will be used to build an algorithm of the Canadian Food Inspection Agency's ERA-Feed Mill model.